Privacy Policy

What the extension stores (on your device)

Everything lives in your browser's local extension storage (chrome.storage.local):

• Your watches — the page URL, what to watch for (your condition), the optional element you picked, and check frequency.
• Normalized text snapshots of the watched content, so we can detect changes. History is capped and old snapshots are discarded.
• Your settings — theme, quiet hours, and any optional integrations you enable.

This data never leaves your device unless you turn on one of the optional features below.

What we never store

• No passwords or credentials. To watch a logged-in page, the extension reuses your existing browser session — it never asks for or stores your login.
• No raw page HTML is retained — only the normalized text needed to diff changes.
• No analytics, no tracking, no advertising identifiers.

Optional cloud AI (off by default)

Matching your natural-language condition runs on your device using deterministic rules and Chrome's built-in AI. For ambiguous cases you may optionally enable a cloud fallback using your own Anthropic (Claude) API key. When — and only when — you enable this:

• Only the changed text delta and your condition are sent to Anthropic to make the yes/no decision — never your whole browsing history.
• The request uses your own API key and is subject to Anthropic's privacy policy.
• You can turn this off at any time; with it off, no page content ever leaves your device.

Optional alert channels (off by default)

You may optionally route match alerts to a webhook URL or a Telegram bot you configure. If you enable these, when a watch matches, the extension sends that destination a short message containing the watch's name, URL, your condition, and a summary of the change. These requests go directly from your browser to the service you chose — we do not proxy or receive them. The values you enter (webhook URL, bot token, chat ID) are stored locally on your device.

Optional Cloud Sync (beta, off by default)

Cloud Sync keeps your watches running on a server you control even when Chrome is closed. There is no shared backend: if you enable it, you supply your own server URL and access token — for example a Cloudflare Worker you self-host from this project's server/ directory. When — and only when — you enable it:

• The extension sends your watch definitions (page URL, your condition, check settings, and any webhook/Telegram values you configured) to your server, and polls it for matches it found while Chrome was closed.
• Data flows only between your browser and the server you control — we never receive it, and the server URL and token are stored locally on your device.
• Turn it off at any time to return to fully on-device checking.

Permissions

The extension requests access to a site only when you start watching it (per-site, at the moment you ask), not to "all your data on all websites." It uses that access to re-fetch the page on your schedule and check your condition. Notifications, alarms, storage, and an offscreen document are used to schedule checks and alert you.

Your control

• Delete any watch (and its snapshots) at any time, or remove the extension to erase everything.
• Export / import your watches as a JSON file from Settings.
• Revoke a site's access from Chrome's extension settings whenever you like.

Contact

Questions or requests? Email notify-me@navbytes.io.